radekosmulski

class ConcatDataset(Dataset):
    """Concatenates a dataset and an iterable of appropriate size."""
    def __init__(self, ds, y2):
        assert(len(ds)==len(y2))
        self.ds,self.y2 = ds,y2
    def __len__(self): return len(self.ds)
    def __getitem__(self, i):
        x,y = self.ds[i]
        return (x, (self.y2[i],y))
    def denorm(self, im): return self.ds.denorm(im)

radekosmulski

import fire
import fastai
from fastai.vision import *
from torch import nn
from fastai.metrics import top_k_accuracy

path = untar_data(URLs.CIFAR)
data = ImageDataBunch.from_folder(path, valid='test')

class block(nn.Module):

radekosmulski

from fastai.vision import *
from fastai.script import *
from torch import nn
from fastai.metrics import top_k_accuracy

path = untar_data(URLs.CIFAR)
data = ImageDataBunch.from_folder(path, valid='test')

class block(nn.Module):
    def __init__(self, n_in, n_out, two_d=True):

radekosmulski

from fastai.vision import *
from fastai.script import *
from torch import nn
from fastai.metrics import top_k_accuracy

path = untar_data(URLs.CIFAR)
data = ImageDataBunch.from_folder(path, valid='test')

class block(nn.Module):
    def __init__(self, n_in, n_out, two_d=True):

radekosmulski

# from the root of fastcore:

python -c "from nbdev.cli import notebook2html; notebook2html()"

gem install jekyll bundler # need to have Ruby installed

cd docs

bundle install

radekosmulski

# if after running the below you still experience issues, take a look at the current
# docker image definitions here: https://github.com/NVIDIA-Merlin/Merlin/tree/main/docker
# (maybe something has changed)

sudo apt update -y --fix-missing
sudo apt install -y --no-install-recommends libexpat1-dev libsasl2-2 libssl-dev graphviz openssl protobuf-compiler software-properties-common
sudo apt autoremove -y
sudo apt clean
sudo rm -rf /var/lib/apt/lists/*

radekosmulski

radekosmulski

Summary of Fastai.ai's in-depth discussion of types of normalization

# simulated batch of images
x = torch.rand(64, 3, 224, 224)
# or some number of layers up the convolutional stack
x = torch.rand(64, 256, 32, 32)

BatchNorm:

radekosmulski

radekosmulski